What question did this study set out to answer?

January 16, 2026

Identification of Risk Factors and Development of Machine Learning Prediction Models for Inpatient Calcium Pyrophosphate Deposition Disease Flares

NPNutnicha PattaravimonpornMahidol University KSKajbundit SurasitBhumibol Adulyadej Hospital PSPanchalee SatpanichNavamindradhiraj University

Key Points

This study aims to define independent risk factors for inpatient calcium pyrophosphate deposition disease flares and to develop predictive models for early management.
Conducted a retrospective case-control study at a tertiary care hospital
Matched adults with CPPD flares to controls without flares
Utilized univariate and multivariate logistic regression analyses
Developed predictive models using decision tree, random forest, logistic regression, and extreme gradient boosting algorithms
Evaluated model performance with precision, recall, accuracy, and AUROC
Identified advanced age, female sex, and in-hospital antibiotic use as independent predictors of CPPD flares
The random forest model achieved the highest accuracy at 0.85
Multivariate analysis provided significant odds ratios for each risk factor

Abstract

Objective: The risk factors for inpatient calcium pyrophosphate deposition disease (CPPD) flares remain poorly defined. This study aimed to identify independent risk factors for inpatient CPPD flares and develop exploratory predictive models to support early recognition and management. Methods: A retrospective case-control study was conducted at a tertiary care hospital from January 2015 to December 2022. Adults aged ≥18 years with confirmed inpatient CPPD flares were matched to controls admitted on the same date and unit without a CPPD flare during hospitalization. Univariate and multivariate logistic regression analyses were used to identify independent risk factors. Exploratory predictive models were developed using decision tree, random forest (RF), logistic regression, and extreme gradient boosting (XGBoost) algorithms. Model performance was evaluated using precision, recall, F1-score, accuracy, and area under the receiver operating characteristic curve. Results: A total of 324 hospitalized patients (162 with CPPD flares and 162 controls) were included. Multivariate analysis demonstrated that advanced age (OR: 1.08; 95% CI: 1.06-1.11; p <0.01), female sex (OR: 1.80; 95% CI: 1.05-3.07; p =0.03), and in-hospital antibiotic use (OR: 1.98; 95% CI: 1.08-3.64; p =0.03) were independent predictors of inpatient CPPD flares. Among the predictive models, the RF model achieved the highest accuracy (0.85) and demonstrated strong discriminative performance (AUROC = 0.89). Conclusions: Advanced age, female sex, and in-hospital antibiotic therapy independently increased the risk of inpatient CPPD flares. The RF model provides a promising proof of concept but requires external validation.

Ask AI

Helpful

Bookmark

View Full Paper